Three Trust Decisions in Full
| Property | Value |
|---|---|
| 📅 Project | OmniRisk · Financial Workflow Dashboard |
| 🔖 Phase | Design |
| 🗂️ Count | 3 trust decisions — the load-bearing design calls |
Three decisions define the design. They are the load-bearing calls that determine whether the design earns adoption from skeptical operators. Each decision resolves a specific tension surfaced in research and synthesis.
Each decision is documented with what was decided, what alternatives were considered, why one choice won, and the research connection that grounds the decision in user evidence. These decisions carry the case study argument — they are the decisions most likely to be probed in an interview.
💡 On supporting design work: The widget-based dashboard architecture, in-table AI highlighting, and automatic audit trail generation are Core features documented on the Feature Set page — they are the structural outputs of the trust decisions rather than co-equal decisions themselves. Every Core feature traces back to one or more of the three decisions below.
The decision: Surface AI confidence scores and reasoning in the AI Recommendation modal — not as columns in the Currency Trade Table.
The alternative considered: Confidence and reasoning columns added directly to the Currency Trade Table so the operator sees everything at once.
Why the modal won: The badges in the table are the triage layer — Repatriate, Monitor, or Hold tells the operator which positions need action. Those badges are triggered by fund-strategy rules set upstream by committee, against the fund's investment strategy. The badge is a trigger to act, not a recommendation to second-guess. The granular reasoning belongs one layer down: when the operator clicks Execute, the modal opens the detail — confidence score, rule attribution, AI-generated rationale, data source attribution, estimated projected impact. The table stays fast for triage; the modal goes deep for the decision the operator is about to authorize.
What the modal actually contains: Confidence score displayed as a band (low / medium / high) with percentage — the confidence-band design surfaced during testing when the same 78% read as "high" to one participant and "moderate" to another; the band resolved that ambiguity. AI-generated reasoning paragraph explaining which rule fired and why. Data source attribution — what data the AI narration is grounded in, what data it never sees. Estimated projected impact. Paired action buttons (Accept AI Recommendation and Manual Override) of equal visibility.
Research connection: The FX trader described wanting to verify AI outputs before acting. The middle-office analyst named full transparency as a condition for trust. The modal serves both without slowing either down at the action layer.
Alternative rejection reasoning: Surfacing confidence and reasoning as inline table columns clutters the triage layer with detail the operator only needs at the decision moment. Triage speed matters more than triage depth — operators scan the table to identify what needs action, then commit at the modal.
The decision: Enforce a sequential progression across the workflow — Fetch FCM → AI Processing → Run Analysis → View Insights — where each phase locks until the prior completes. Three user-clickable CTAs (Fetch FCM, Run Analysis, View Insights) with AI Processing as an automatic phase between them.
The alternative considered: Free navigation across all workflow phases at any time, with the user managing sequence dependencies themselves.